By Charles Skorina
Chief investment officers at endowments, foundations, and family offices are the top guns of the institutional investment world. They have an infinite investment horizon, a global playing field, and can invest in anything anywhere - within the broad policy limits set by their institution.
We, and many others, regard the CIOs at major American universities and foundations as the best of the best.
We lean heavily on university endowments for our performance studies and benchmarks because that's where the data is.
Foundations, family offices, and Wall Street firms employ top investment professionals, but it's difficult to extract meaningful data from opaque sources. So, we go with what we can get.
A five-year return: the Goldilocks number
Institutions go on forever, but chief investment officers unfortunately don't.
Five-year returns give us a good -- albeit imperfect - picture of how CIOs are doing their jobs. A longer timeframe would blur the responsibility for results as CIOs come and go.
In our chart below, the median tenure of CIOs happens to be exactly 5 years. (The mean is higher, tipped by a handful of very senior CIOs.)
In our SEER reports (aka: Skorina's Enhanced Endowment Reports) we use the five-year rankings for our own headhunting purposes, and we let you look over our shoulders.
Boards and investment committees set broad policies. Executing those policies in the day-to-day scrum of the markets -- especially in the hiring, firing and monitoring of external managers -- is the province of the CIO and his/her staff. As recruiters, we try to understand who's doing it well, or not so well.
We know that nominal returns don't reflect the different risk-appetites of different investors; and ranking doesn't tell the whole story.
There may be good reasons why one institution prefers a more, or less, conservative risk-return trade-off versus its peers. We'll say more on that point down below. But five-year nominal return is our starting point.
Our dataset consists of 34 big, over $1 billion AUM, North American endowments reporting as of early December.
That's only about a third of the whole big-endowment roster. It leaves about 50 who have yet to be heard from, and another dozen or so who disdain to report their returns at all, even when we ask politely.
But our gang of 34 is big enough to show us how the whole league has performed and it includes many of the brand-name schools and all the traditional Ivys.
Another demi-decade bites the dust
Return with us now to those bright, beckoning days of July 2014. How young and hopeful we all looked!
That was already the fifth year of the same post-Crisis bull market we're still riding.
We're holding up pretty well; the spavined bulls not so much.
As the fiscal year ended in June the S&P 500 had climbed from 1,961 to 2,716: up 39 percent. On a total return basis, it gave investors 13.4 percent annually.
U.S. bond investors were handed annual total returns of about 2.3 percent. So, a domestic bond portfolio would have grown about 12 percent over five years with interest reinvested.
A 60/40 stock/bond portfolio, then, would have returned about 9 percent compounded, by our calculation.
There were the usual alarums and excursions along the way (and endowments these days have to wrangle much more exotic creatures than stocks and bonds).
But, those numbers give us a broad sense of how the world has treated investors since 2014.
Here's how it worked out for our 34 big endowments.
|
|
Yo, Canada!
Our 34 endowments returned an average of 9.1 percent over five years, unweighted.
We think that's a good predictor of NACUBO's official average for all big endowments, which won't be out for some weeks.
[See our prior newsletter for chief investment officers pay projections in 2018]
Up on the top half of the list we see the usual suspects, and a few not so usual.
One persistent overachiever is Paula Volent at little Bowdoin College in Maine.
Her pool is one-twentieth the size of Yale's. But she surpasses her old mentor David Swensen (again) and almost matches MIT's Seth Alexander, who leads the league with 12 percent annualized return for five years.
All four of our top performers are Yalies: Mr. Swensen himself and three alumni of the Yale Investments Office.
Also, among the leaders is Ms. Gitta Kulczycki (pronounced KUL-CHIKI, we think). Although she did not attend Yale, she's doing a fine job at the University of Alberta endowment out on the Canadian prairie.
Her pool is even smaller than Bowdoin's, but she and her Canadian colleagues - Ms. Leblanc at McGill and Mr. Silgardo at University of British Columbia - seem to be out-investing most of the Ivy League. Who knew?
Adjusting for risk: Do we need a Sharper Image?
Using a 5-year nominal return reduces the statistical noise, and we think that's a step forward in evaluating performance.
But that ranking still seems to imply that a higher nominal return is always better.
We know that's not necessarily true, because investors have different risk-appetites.
If endowments routinely published their Sharpe ratios we could also rank on that basis. That could give us a deeper understanding of what they're up to. But they don't so, we can't.
But we can offer some anecdotal evidence to remind us that nominal returns aren't the whole story.
Mark Schmid's 5-year return at University of Chicago, for instance, puts him at the bottom of our list. But he has been signaling for years that he and his board are following a lower-risk policy that suits their needs, but which won't win the nominal-return derby.
We haven't discussed this topic directly with Mr. Schmid, but we did get an intriguing heads-up from an analyst at one major consulting firm who's privy to a lot of data.
Our contact, of course, can't disclose numbers they've obtained in confidence from their clients. But, looking at our data, he confirmed that, on a risk-adjusted basis, Mr. Schmid's 5-year returns would be much closer to the top of our list than the bottom.
Then, there's Mr. Zimmerman, ex-CIO at UTIMCO, who wrote in 2013:
...the Endowment's investment returns have lagged other large endowments primarily due to the Endowment's lower risk profile...While it is the case that risk has been rewarded over the past few years, there is agreement [among staff, Board, and Regents] that the necessity to protect principal supersedes the desire for higher investment returns."
In our current chart, UTIMCO is about two-thirds down the list, which has been typical for them.
There's been turnover at the UTIMCO board, and Britt Harris got the CIO job last year, following Mr. Zimmerman's departure. And there has been speculation about a policy shift in Austin that would raise returns, although we haven't seen much evidence of that yet.
When we read the latest report from CIO Jeremey Crigler at Tulane University, we noted that he was stoked about TU's recent risk-adjusted performance, and he cited his Sharpe Ratio to prove it.
Cambridge Associates is their general consultant and we assume they did the calculations and provided peer data to show that TU did very well on a risk-adjusted basis.
He wrote:
Generally, strong returns should be viewed with skepticism since higher returns are often the result of taking additional risk...[but], the risk of the Endowment [over 5 years 2014-2018], measured by the standard deviation of returns, was just 4.1%, ranking in the 5th percentile. As a result, the Endowment's Sharpe ratio, or the return per unit of risk, was an extraordinary 2.1x, ranking 6th among peers.
In our chart, we see that TU's respectable 9.4 percent nominal return puts them right in the middle of our chart, beating half of the vaunted Ivy League. Not too shabby.
What if we had Sharpe numbers for all 34 of our CIOs? How would Tulane's 2.1 SR number stack up? We suspect that it would move Tulane (along with Chicago and UTIMCO) higher in the rankings. But, without more data, we're just speculating.
Processing the Markov numbers
We can't offer you a list ranked by Sharpe Ratio, but a firm called Markov Processes is working on that issue. They have a proprietary methodology which is supposed to estimate risk-adjusted returns even when funds aren't very transparent about the relevant data.
(The name of the firm is a math-geek joke, by the way. The Russian mathematician Andrey Markov is famous for inventing a statistical gadget called a Markov Process. Mr. Michael Markov, founder of MPI, is presumably no relation. But, Prof. A. Markov, who died in 1922, definitely had a more impressive beard.)
MPI didn't analyze all the big endowments, just the eight Ivys. You can read their white paper here:
Their report breaks down into two parts:
First, they at look at the 10 years of publicly-available nominal returns. No risk-adjustment involved.
Second, they use their proprietary method to estimate standard deviations and Sharpe ratios for the eight Ivys in the same period.
It was that first part that got some media attention. The headline was that none of the Ivys beat a 60/40 blend for the 2009-2018 decade.
It even got coverage from James B. Stewart, a veteran feature writer at the New York Times, who covered the story last Friday.
Our friend Leo Kolivakis at the PensionPulse blog also weighed in. He solicited some very insightful comments from three industry heavyweights, including Leo De Bever, former CEO of AIMCO in Alberta. They were interested, but a bit skeptical about MPI's research.
MPI's 10-year lookback period means that their numbers don't neatly mesh with our 5-year chart, which says that most Ivys did beat the 60/40 portfolio in recent years.
But, in any case, we're also slightly skeptical about their headline claim.
We think the 60/40 return number they use for this period is a little high.
And, again, if you break the 10-year period into two 5-year chunks (2009-2013, and 2014-2018) then, as we see in out chart above, most of the Ivys (6 out of 8) did beat the 60/40 in the later, more recent period.
To be clear, we don't think MPI was deliberately misleading in their report. But some of the media coverage it generated was a little overwrought.
(We say a more about these technical issues in an appendix down below.)
Markovs' secret sauce
That leaves the second part of the MPI report. Here, they use their secret-sauce methodology to conclude that even risk-adjustment doesn't help the hapless Ivys beat the 60/40.
Using Sharpe ratios, the gap is even starker. A 60/40 portfolio is not only simpler and cheaper, but also looks less volatile in terms of standard deviation over ten years.
This little table summarizes the MPI numbers
Markov Processes International
Risk and Return: Ivys vs 60/40
|
|
|
NA: for standard deviation, lower is better; for Sharpe ratio, higher is better.
On our nominal-return rankings up above, Princeton, Yale, and Dartmouth all look better than Penn among the Ivys.
But, on MPI's risk-adjusted basis, Penn rises to the top, while the others sink down to the middle of the list.
Alas, risk adjustment doesn't help Cornell or Harvard. They finish at the bottom on the MPI list, just as they do on ours.
Modeling the risk of the exotic alternative investments in modern portfolios is no easy thing, even for the insiders. But if MPI can pull it off, it could be very useful to outside analysts of all sorts of non-transparent funds.
MPI also published a deep-dive, academic paper earlier this year which goes into more detail about their methodology: Fragiskos, Ryan and Markov: Alpha and performance efficiency of Ivy League endowments (2018)
Remembering real-world risk
All this talk of Sharpe ratios and standard deviations seems rather theoretical. Let's take a more concrete perspective.
Most of our readers still have a vivid recollection of the stomach-churning events of 2008/2009 as the financial meltdown crushed returns for both institutions and retail investors. It doesn't feel academic when your retirement is evaporating.
In FY 2009, Harvard and Yale reported negative 30 and 29 percent returns, respectively. But Penn's endowment, then headed by Kristin Gilbertson, lost much less: only 17 percent. That is the practical, real-world consequence of Penn's higher Sharpe ratio on that chart.
Despite her good results in 2009, Ms. Gilbertson left after the departure of trustee Howard Marks -- BA Wharton, MBA U Chicago, and co-founder of Oaktree Capital Management -- joining the family office of Len Blavatnick at Access Industries as chief investment officer.
Penn is doing fine under successor Peter Ammons (they beat Yale this year and placed third among the Ivys on a 1-year basis). And, according to the little MPI chart above, Penn is still handily leading the Ivys on a risk-adjusted basis
Appendix
We had some questions about the MPI white paper referenced above.
We think the 10-year average return they used for a 60/40 index is on the high side.
The components are, of course, the S&P 500 Total Return (with reinvested dividends) and the Bloomberg Barclay's Aggregate Bond index.
The 60/40 is just a weighted average of those two. And the indexes themselves are published every trading day by their respective owners. The data is all public, so how could there be different numbers?
Well, we suspect that the very competent quants at MPI might have used daily numbers for both indexes. Because, why not use the biggest dataset? That would give them about 2,500 daily observations for each index.
Looking at the day-to-day changes and annualizing them, one could get a number that's a little higher than if you took a less-fancy approach, using just ten annual observations or, perhaps 40 quarterly observations.
Generally, when you use more observations and smaller intervals in this kind of calculation, you bias the annualized return upward.
When banks and asset managers published their Q2 reviews they routinely reported historical returns for major indexes. The number they printed for 10-yr return on the S&P and the Agg, were 10.20 percent and 3.70 percent, respectively.
When NACUBO gets around to publishing their public tables for FY2018, they will undoubtedly print the same numbers for those indexes, cheek-by-jowl with their historical endowment returns. The clear implication is that the former are useful benchmarks for understanding the latter, and that they are being presented on a comparable basis.
If you use those conventional numbers to calculate the 60/40 weighted average for five years, you get 7.59, not 8.1.
It's a small difference, but when using 7.6, three Ivys beat the 60/40 over 10 years. Using 8.1 means that none do.
We think using annual observations, or at least quarterly observations, is more real-world. We don't think managers rebalance every day.
Second: the 10-year period 2009-2018 is historically unusual. In most 10-year intervals since 2003, big endowments did much better than the 60/40.
The Ivys, while catnip to the media, are less than ten percent of the whole universe of big endowments.
We constructed rolling 10-year averages for the over-$1 billion endowments and the 60/40 and found that the average big endowment beat the 60/40 in 12 out of 15 periods from 2003 to 2018. The 60/40 looked better in the three most recent 10-year windows.
There's also some indication that the Ivys-vs-60/40 shortfall maxed out in the 5 years 2009-2013 and that the shortfall was much less evident in the more recent five years 2014-2018, which happens to correspond to the five years in our chart up above. Charles Skorina works with leaders of endowments, foundations, and institutional asset managers to recruit Board Members, Executives Officers, Chief Investment Officers and Fund Managers. Mr. Skorina also publishes The Skorina Letter, a widely read professional publication providing news, research and analysis on institutional asset managers and tax-exempt funds. Prior to founding CASCo, Mr. Skorina worked for JP MorganChase in New York City and Chicago and for Ernst & Young in Washington, D.C. |